Spark 之 DataFrame

创建DataFrame

val df = Seq(
  (1, 100, 10),
  (2, 100, 12),
  (3, 200, 10),
  (4, 200, 12),
  (5, 300, 12),
  (6, 300, 15)
).toDF("StudentID", "Class", "Age")

df.createOrReplaceTempView("student")
spark.sql("SELECT * FROM student").show
+---------+-----+---+
|StudentID|Class|Age|
+---------+-----+---+
|        1|  100| 10|
|        2|  100| 12|
|        3|  200| 10|
|        4|  200| 12|
|        5|  300| 12|
|        6|  300| 15|
+---------+-----+---+

spark.sql("SELECT min(Class), max(Class) FROM student group by Class").show
+----------+----------+
|min(Class)|max(Class)|
+----------+----------+
|